Publication number | US6351491 B1 |

Publication type | Grant |

Application number | US 09/594,401 |

Publication date | Feb 26, 2002 |

Filing date | Jun 15, 2000 |

Priority date | Jun 23, 1999 |

Fee status | Lapsed |

Also published as | WO2000079802A1 |

Publication number | 09594401, 594401, US 6351491 B1, US 6351491B1, US-B1-6351491, US6351491 B1, US6351491B1 |

Inventors | Hung-Ju Lee, Tihao Chiang, Iraj Sodagar, Ya-Qin Zhang |

Original Assignee | Sarnoff Corporation |

Export Citation | BiBTeX, EndNote, RefMan |

Patent Citations (9), Non-Patent Citations (5), Referenced by (87), Classifications (24), Legal Events (6) | |

External Links: USPTO, USPTO Assignment, Espacenet | |

US 6351491 B1

Abstract

A method and apparatus for recursively optimizing the rate control of a hierarchical subband coding system that offers spatial, quality and/or complexity scalabilities. The rate control method recursively adjusts the quantizer scale for each layer of a subband tree, i.e., a subband decomposed image.

Claims(20)

1. Method for encoding an input image into a bitstream, said method comprising the steps of:

(a) generating a subband tree for the input image, where said subband tree has a plurality of subband layers of coefficients;

(b) quantizing said plurality of layers of coefficients with a plurality of quantizer scales, where each quantizer scale corresponds to a layer of said subband tree;

(c) recursively adjusting at least one of said plurality of quantizer scales in accordance with a quantization error that is computed from at least two of said plurality of subband layers; and

(d) encoding said quantized coefficients.

2. The method of claim 1 , wherein said quantization error of adjusting step (c) is computed in accordance with a measure of mean square quantization error.

3. The method of claim 1 , wherein said subband tree is a wavelet tree.

4. The method of claim 3 , wherein said quantizing step b) comprises the steps of:

(b1) computing a target bitrate for each of said plurality of subband layers by using a polynomial model; and

(b2) quantizing coefficients from each of said plurality of subband layers in accordance with said computed target bitrate for each of said plurality of subband layers.

5. The method of claim 4 , wherein said polynomial model is expressed as: $\frac{{R}_{i}-{H}_{i}}{{S}_{i}}=\frac{{A}_{1}}{{Q}_{i}}+\frac{{A}_{2}}{{\left({Q}_{i}\right)}^{2}}$

where R_{i }is a target bitrate for encoding a subband layer i, H_{i }is a number of bits for encoding header information, S_{i }is a measure of said quantization error in a subband layer i, Q_{i }is a quantization scale for a subband layer i, and A_{1 }and A_{2 }are first order and second order coefficients.

6. The method of claim 3 , wherein said adjusting step c) comprises the step of:

(c1) computing said quantization error in accordance with a normalized mean square quantization error.

7. The method of claim 6 , wherein said normalized mean square quantization error is expressed as: $\mathrm{NMSE}\ue8a0\left(i\right)=\left(\sum _{j}\ue89e\mathrm{MSE}\ue8a0\left(j\right)+\left(1\ue2fb\left(k-i-1\right)\right)\right)\ue2f7\left(k-i\right),$

where NMSE(i) is said normalized mean square quantization error for a subband layer i, k is a number of said subband layers, j denotes a spatial location in said subband layer i and symbols >> and << are right and left bit shifting operations, respectively.

8. The method of claim 6 , wherein said adjusting step c) further comprises the step of:

(c2) computing a new target bitrate for at least one of said plurality of subband layers in accordance with a weighting factor that accounts for quantization error changes in a next iteration of coding.

9. The method of claim 8 , wherein said encoding step (d) encodes said quantized coefficients in accordance with said new target bitrate.

10. The method of claim 8 , wherein said weighting factor is expressed as:

where actual_bitrate is a bitrate resulting from actual encoding, target_bitrate is a target bitrate, layer_MSE[i] is a mean square quantization error for a subband layer i, current_QP is a current quantization scale, total_MSE is a sum of said mean square quantization errors for all of said plurality of subband layers, and weight[i] is said weighting factor for a subband layer i.

11. The method of claim 8 , wherein said new target bitrate is expressed as:

where R[i] is said new target bitrate for a subband layer i, target_bitrate is a target bitrate for a current frame, current_MSE[i] is a current mean square quantization error for a subband layer i, weight[i] is said weighting factor for a subband layer i.

12. The method of claim 4 , wherein parameters of said polynomial model are updated using actual encoding results.

13. The method of claim 1 , wherein said encoding step (d) encodes said quantized coefficients in accordance with multiscale zerotree entropy encoding (MZTE).

14. Apparatus for encoding an input image into a bitstream, said apparatus comprising:

a subband tree generator for generating a subband tree for the input image, where said subband tree has a plurality of subband layers of coefficients;

a quantizer, coupled to said subband tree generator, for quantizing said plurality of layers of coefficients with a plurality of quantizer scales, where each quantizer scale corresponds to a layer of said subband tree;

a rate controller, coupled to said quantizer, for recursively adjusting at least one of said plurality of quantizer scales in accordance with a quantization error that is computed from at least two of said plurality of subband layers; and

an encoder, coupled to said quantizer, for encoding said quantized coefficients.

15. The apparatus of claim 14 , wherein said quantizer computes said quantization error in accordance with a measure of mean square quantization error.

16. The apparatus of claim 14 , wherein said rate controller computes a target bitrate for each of said plurality of subband layers by using a polynomial model, and said quantizer quantizes coefficients from each of said plurality of subband layers in accordance with said computed target bitrate for each of said plurality of subband layers.

17. The apparatus of claim 16 , wherein said rate controller further computes a new target bitrate for at least one of said plurality of subband layers in accordance with a weighting factor that accounts for quantization error changes in a next iteration of coding.

18. The apparatus of claim 14 , further comprising a symbol assignor, coupled to said encoder, that assigns symbols to said quantized coefficients in accordance with multiscale zerotree entropy encoding (MZTE).

19. A computer-readable medium having stored thereon a plurality of instructions, the plurality of instructions including instructions which, when executed by a processor, cause the processor to perform the steps comprising of:

(a) generating a subband tree for the input image, where said subband tree has a plurality of subband layers of coefficients;

(b) quantizing said plurality of layers of coefficients with a plurality of quantizer scales, where each quantizer scale corresponds to a layer of said subband tree;

(c) recursively adjusting at least one of said plurality of quantizer scales in accordance with a quantization error that is computed from at least two of said plurality of subband layers; and

(d) encoding said quantized coefficients.

20. An encoding system comprising:

a motion compensator for generating a predicted image of a current input frame;

a subband coder, coupled to said motion compensator, for applying a subband decomposition to a difference signal between the input frame and said predicted image, where said decomposition produces a subband tree having a plurality of subband layers of coefficients, and for quantizing said plurality of layers of coefficients with a plurality of quantizer scales, where each quantizer scale corresponds to a layer of said subband tree; and

a rate controller, coupled to said subband coder, for recursively adjusting at least one of said plurality of quantizer scales in accordance with a quantization error that is computed from at least two of said plurality of subband layers.

Description

This application claims the benefit of U.S. Provisional Application No. 60/140,545 filed Jun. 23, 1999, which is herein incorporated by reference.

The present invention relates to an apparatus and concomitant method for optimizing the coding of an image sequence, e.g., motion video. More particularly, this invention relates to a method and apparatus that recursively optimizes the rate control, e.g., adjusting the quantizer scale, for each layer of a hierarchical subband coding system, e.g., a wavelet transform based coding system.

International committees have established various standards for coding/decoding strategies. Examples of such standards include, but are not limited to, the Moving Picture Experts Group Standards (e.g., MPEG-1 (11172-*), MPEG-2 (13818-*) and MPEG-4 (14496-*)), H.261 and H.263. Although the present invention is described below in terms of an MPEG encoding system, it should be noted that the present invention can be adapted to other encoding systems that are compliant to other standards.

Although the MPEG standards specify a general coding methodology and syntax for generating an MPEG compliant bitstream, many variations are permitted to accommodate a plurality of different applications and services such as desktop video publishing, video conferencing, digital storage media and television broadcast. In the area of rate control, MPEG does not define a specific method for controlling the bit rate of an encoder. It is the task of the encoder designer to devise a rate control process for controlling the bit rate such that the decoder input buffer neither overflows nor underflows.

Examples of rate controls for encoding systems are disclosed in U.S. patent application Ser. No. 08/738,228 filed Oct. 24, 1996 and Ser. No. 09/022,349 filed Feb. 11, 1998, which are commonly owned by the Assignee and are herein incorporated by reference. These encoding systems employ rate control methods that compute quantization scale or quantizer scale in accordance with a complexity model having a polynomial form. These rate control methods have been found to be very effective.

However, as new encoding systems are developed or functionalities are added, the rate control methods may have to be adapted or refined. For example, wavelet transforms, otherwise known broadly as an example of a hierarchical subband decomposition, have recently been used for low bit rate image compression because such decomposition leads to a hierarchical multi-scale representation of the source image. The hierarchical multi-scale representation provides a flexible platform for implementing various innovative processing methods.

For example, U.S. Pat. No. 5,412,741 issued May 2, 1995 and herein incorporated by reference discloses an apparatus and method for encoding information with a high degree of compression. The apparatus uses zerotree coding of wavelet coefficients in a much more efficient manner than any previous techniques. The methods as discussed in the '741 patent are known as the “Embedded Zerotree Wavelet” (EZW) method.

Another example is the concept of “multiscale zerotree entropy encoding” (MZTE) which is the subject of U.S. patent application Ser. No. 09/107,544, filed Jun. 30, 1998 which is commonly owned by the Assignee and is herein incorporated by reference. MZTE extends the zerotree entropy (ZTE) encoding method to achieve a fully scalable coding method by implementing a multiscale zerotree coding method. Both ZTE and MZTE coding techniques use the concept of zerotree symbol to achieve the compression gain. Each quantized wavelet coefficient in a wavelet tree is assigned with one of the four symbols: zerotree root (ZTR), valued zerotree root (VZTR), value (VAL) or isolated zero (IZ). A zerotree root denotes a coefficient that is the root of a zerotree. Zerotrees do not need to be scanned further because it is known that all coefficients in such a tree have amplitudes of zero. A valued zerotree root is a node where the coefficient has nonzero amplitude, and all four children are zerotree roots. A VAL symbol identifies a coefficient with a value, either zero or non-zero, but also with some descendant somewhere further along the tree that has a non-zero value. An IZ symbol identifies a coefficient with zero value, but with some descendant somewhere further along with a nonzero.

The MZTE technology has been adopted by the MPEG-4 committee as the Visual Texture Coding tool (VTC) for compression of natural and synthetic images. The VTC coding technique provides useful functionality such as spatial and quality scalabilities, error resilience, and selective region of interests with better image quality. However, without an efficient and effective rate control method, multimedia applications using the MPEG-4 VTC may be limited.

Therefore, there is a need in the art for a method and apparatus that recursively optimizes the rate control, e.g., adjusting the quantizer scale, for each layer of a hierarchical subband coding system that offers spatial, quality and/or complexity scalabilities.

In one embodiment of the present invention, a method and apparatus is disclosed that recursively optimizes the rate control, e.g., adjusting the quantizer scale, for each layer of a hierarchical subband coding system that offers spatial, quality and/or complexity scalabilities. The present invention recursively adjusts each quantization scale of a subband layer in accordance with a weighting factor that prioritizes the importance among the subband or resolution layers.

More specifically, the quantization scale for each subband layer can be recursively adjusted in accordance with a quantization error that is computed from at least two of said plurality of subband layers. Thus, the selection of the quantization scale for each subband layer accounts for the dependency that exists between coefficients of different subband layers. In one embodiment, the measure of the quantization error is a mean square quantization error (MSE), where the weighting factor is determined by the MSE of the next iteration in each individual subband layer. With this newly computed MSE, the weighting factor can be used to forward predict and heuristically reduce the overall MSE for the entire frame.

The teachings of the present invention can be readily understood by considering the following detailed description in conjunction with the accompanying drawings, in which:

FIG. 1 depicts a block diagram of a hierarchical subband encoding system, e.g., a wavelet-based encoding system, incorporating the present invention;

FIG. 2 depicts a block diagram of a hierarchical subband encoder, e.g., a wavelet-based encoder of the present invention;

FIG. 3 is a graphical representation of a wavelet tree;

FIG. 4 is a schematic illustration of the dependency of zerotree based coding;

FIG. 5 depicts a block diagram of a bitstream with layers of spatial resolution scalability;

FIG. 6 depicts a block diagram of a bitstream with layers of SNR or quality scalability;

FIG. 7 depicts a block diagram of a bitstream with combined SNR-spatial scalabilities;

FIG. 8 illustrates a block diagram of a portion of an encoder for generating SNR scalability layers;

FIG. 9 illustrates a flowchart for deriving the optimal quantizer scale for each layer of a hierarchical subband coding system; and

FIG. 10 illustrates an encoding system of the present invention.

To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures.

The present invention is a method and apparatus that recursively optimizes the rate control, e.g., adjusting the quantizer scale, for each layer of a hierarchical subband coding system that may offer spatial, quality and/or complexity scalabilities. Specifically, the present rate control invention is capable of addressing the dependency that is inherent in hierarchical subband coding system. Although the present invention is described below in terms of wavelet decomposition, it should be understood that the present invention is not so limited, and can be broadly adapted to hierarchical subband decomposition in general. Namely, wavelet is a particular type of subband decomposition.

In an ZTE wavelet-based coding system, dependencies exist among zerotree symbols and wavelet layers. A quantization parameter selected for one wavelet layer will interact with other wavelet layers. An effective rate control invention must properly address this dependency criticality.

To illustrate, FIG. 4 is a schematic illustration of the dependency of zerotree based coding. In ZTE, each quantized wavelet coefficient is assigned one of four zerotree symbols: ZTR, VZTR, VAL and IZ. Assuming an image is decomposed into four wavelet decomposition layers, layer-0, layer-1, layer-2 and layer-3, where layer 0 is the coarsest layer (i.e., closest to the DC layer) and layer 3 is the finest layer. Let Q**0**, Q**1**, Q**2** and Q**3** be the quantization parameters or scales for layer-0, -1, -2, and -3 respectively. Then the change of Q**2** value may affect the zerotree symbol assignment of nodes in layer-1 and layer-0. Similarly, the change of Q**1** value may affect the zerotree symbol assignment of nodes in layer-0, but not layer-2. FIG. 4 illustrates such a scenario where without loss of generality, assuming that at iteration i, all of nodes X in layer-0, Y in layer-1 and Z in layer-2 are quantized to 0 with zerotree symbol ZTRs. At the next iteration i+1, Q**2** decreases so that node Z is quantized to non-zero value (i.e., 4 in this case), then node Z is marked as VZTR, then node Y will be marked as IZ since one of its children is not zero. Similarly, node X is marked as IZ. From the above scenario, it becomes clear that the quantization parameter change in a lower layer affects the bitrate used in its upper layer since the statistics of zerotree symbols are changed. The present rate control invention is capable of addressing such dependency.

To better understand the present invention, a description is now provided of the hierarchical subband coding system in which the present rate control method is deployed. FIG. 1 depicts an encoder or portion of an encoding system **100** of the present invention. The encoder **100** contains a block motion compensator (BMC) and motion vector module **104**, a subtractor **102**, a hierarchical subband coder, e.g., a discrete wavelet transform (DWT) coder **106**, a bit rate controller **110**, a DWT decoder **112** and an output buffer **114**. It should be noted that the terms “coder” and “encoder” are used interchangeably in the specification.

In general, the input signal is an image (a two-dimensional array of pixels (pels) defining a frame in an image sequence). To accurately transmit the image through a channel, the spatial and temporal redundancy in the image frame sequence must be substantially reduced. This is generally accomplished by coding and transmitting only the differences between successive frames. In brief, the encoder **100** performs three broad functions: first, it produces a plurality of motion vectors that represent motion that occurs between frames; second, it predicts the present frame using a reconstructed version of the previous frame combined with the motion vectors; and third, the predicted frame is subtracted from the present frame to produce a frame of residuals that are coded and transmitted along with the motion vectors to a receiver. Within the receiver, a decoder reconstructs each video frame using the coded residuals and motion vectors. It should be noted that certain anchor frames, e.g., I frames, may be coded without performing motion estimation.

More specifically, FIG. 2 depicts a block diagram of the encoder **106** of the present invention. The encoder **106** contains a wavelet tree generator **204**, an optional wavelet tree reorganizer **208**, a quantizer **210**, a symbol assignor **212**, and an entropy encoder **214**. Each of these components is connected in series to process an image at port **202** into a coded output image at port **216**. The input image is typically a pixelated (digitized) photographic image as can be produced from an image scanner or a computer graphics system. However, the input image can also be a frame within a series of frames of video images or a motion compensated residual frame produced by a video encoding system. In general, the invention processes any form of digitized image or portion thereof, i.e., any form of two-dimensional data.

The wavelet tree generator **204** performs a wavelet hierarchical subband decomposition to produce a conventional wavelet tree representation of the input image. To accomplish such image decomposition, the image is decomposed using times two subsampling in each of two-dimensions into high horizontal-high vertical (HH), high horizontal-low vertical (HL), low horizontal-high vertical (LH), and low horizontal-low vertical (LL), frequency subbands. The LL subband is then further subsampled times two in each of two dimensions to produce a set of HH, HL, LH and LL subbands. This subsampling is accomplished recursively to produce an array of subbands such as that illustrated in FIG. 3 where three subsamplings have been used. Preferably four or more subsamplings are used in practice, but the present invention can be adapted to any number of subsamplings. The parent-child dependencies between subbands are illustrated as arrows pointing from the subband of the parent nodes to the subbands of the child nodes. The lowest frequency subband is the top left LL_{3}, and the highest frequency subband is at the bottom right HH_{1}. In this example, all child nodes have one parent. A detailed discussion of subband decomposition is presented in J. M. Shapiro, “Embedded Image Coding Using Zerotrees of Wavelet Coefficients”, IEEE Trans. on Signal Processing, Vol. 41, No. 12, pp. 3445-62, December 1993.

Returning to FIG. 2, the quantizer **210** quantizes the coefficients of the wavelet tree via path **206** in a “depth-first” pattern. Namely, the method traverses each tree from the root in the low-low subband (LL_{3}) through the children. Although the foregoing description of the depth-first scanning pattern was discussed as a “top down” pattern, a depth-first scanning pattern also includes scanning from the bottom up. As such, the quantization can also be accomplished by starting at a tree's “leaves” (the bottom-most nodes) and proceeding up the tree. Once that tree was complete, the quantization process would quantize another tree, and another, and so on until all the nodes in all the trees were quantized. As shall be discussed below, the quantization scale selected for each layer will be recursively adjusted to optimize the rate control.

After quantization, at each node of the tree, the quantized coefficient has either a zero value or a non-zero value. The quantized coefficients of the wavelet tree are efficiently encoded by again scanning each tree. Thus, symbol assignor **212** operates by traversing the tree and assigning particular symbols to each node depending upon the node's quantized value as well as the quantized values of each node's descendants.

In turn, the symbols and values are encoded using an entropy coder **214**, such as a conventional arithmetic coder. Although the encoder **106** is described as comprising various elements, those skilled in the art will realize that these elements or the functions performed by these elements can be implemented within or without the encoder **106**, e.g., outside of the encoder **106**, but within the encoding system **100** of FIG. **1**.

In the present invention, the above ZTE method is adapted to encode wavelet trees to generate bitstreams with flexible degrees of spatial, quality and complexity scalabilities. More specifically, FIGS. 5-7 illustrate three different examples of bitstreams having different scalabilities.

FIG. 5 illustrates a block diagram of a bitstream **500** with M layers of spatial resolution scalability. Namely, the bitstream is constructed such that the information representing spatial resolutions **512**-**542** of an input image corresponds to different portions **510**-**540** of the bitstream **500**. In this fashion, if a decoder needs to obtain a spatial resolution **512** of the input image, then the decoder simply decodes the corresponding portion **510** of the bitstream. Thus, if a decoder needs to obtain higher spatial resolutions of the input image, the decoder simply decodes the relevant portions of the bitstream as needed.

FIG. 6 illustrates a block diagram of a bitstream **600** with N layers of SNR or quality scalability. Namely, the bitstream is constructed such that the information representing different qualities **612**-**642** of an input image corresponds to different portions **610**-**640** of the bitstream **600**. In this fashion, if a decoder needs to obtain a particular quality **612** of the input image, then the decoder simply decodes the corresponding portion **610** of the bitstream. Thus, if a decoder needs to obtain higher qualities of the input image, the decoder simply decodes the relevant portions of the bitstream as needed.

Finally, in FIG. 7, the bitstream **700** has M layers of spatial resolution and N layers of SNR scalability, i.e., combined SNR-spatial scalabilities. Namely, the bitstream is constructed such that the information representing different combined SNR-spatial scalabilities **712**-**742** of an input image corresponds to different portions **710**-**740** of the bitstream **700**. In this fashion, if a decoder needs to obtain a particular combination of SNR-spatial scalability **712** of the input image, then the decoder simply decodes the corresponding portion **710** of the bitstream. The number and the kind of scalability (SNR, spatial) are described in the bitstream by the encoder.

FIG. 8 illustrates a block diagram of a portion **800** of an encoder for generating SNR layers. More specifically, encoding portion **800** of the encoder comprises a plurality of stages **810** _{1=n}, where each stage assists in generating a SNR layer.

In operation, the wavelet coefficients of the input image of a particular spatial resolution (different spatial resolutions of the input image can be used) are quantized with the quantizer Q_{0 } **812**. These quantized coefficients are scanned by ZTS module **814** using the above zerotree concept and then the significant maps and quantized coefficients are entropy coded by entropy (or arithmetic) coder **816** as discussed above. The output of the entropy coder **816** at this level, BS**0**, is the first portion of the bitstream, e.g., the first SNR layer.

Next, the quantized wavelet coefficients of the first layer are also reconstructed and subtracted from the original wavelet coefficients via inverse quantizer **818** and buffer **819**. The “residual wavelet coefficients” are then fed into the second stage **810** _{2 }of the coder in which the wavelet coefficients are quantized with Q_{1}, and then zerotree scanned via ZTS module **824** and entropy coded via entropy coder **826**. It should be noted that the residual wavelet coefficients may represent the error introduced by the quantization process. As such, subsequent outputs BS_{x }can be perceived as “refinements” that can be used by the decoder to refine the reconstruction of the quantized wavelet coefficients. However, it should also be noted that changing the quantizer scale in the next stage may also introduce new wavelet coefficients that may not have existed at the above stage (e.g., these new wavelet coefficients were previously quantized to zeros). The quantization process of the present invention is further described below. The output of this second stage, BS**1**, is the second portion of the output bitstream, e.g., the additional information that when combined with the first SNR layer produces the second SNR layer.

The quantized coefficients of the second stage **810** _{2 }are also reconstructed and subtracted from the original coefficients, where the process is continued for the next stage and so on. As shown in FIG. 8, N+1 stages of the coder provides N+1 layers of SNR scalability. Each stage presents one layer of SNR. To obtain spatial (or both spatial and SNR) scalability, different spatial resolutions of the input image can be forwarded as input on path **805**. For example, a plurality of different spatial resolutions of the input image can be processed by the first stage **810** _{1 }to generate a plurality of spatial scalability. In turn, if both spatial and SNR scalability is desired, each spatial resolutions of the input image can be processed by subsequent stages of the encoder portion **800**. Thus, the MZTE method provides a very flexible framework to support the right tradeoff between layers and types of scalability, complexity and coding efficiency for any multimedia application.

FIG. 9 illustrates a flowchart of a method **900** for deriving the optimal quantizer scale for each layer of a hierarchical subband coding system. It should be noted that a “subband layer” is defined broadly to include a spatial and/or quality layer of a hierarchically subband decomposed image. Although wavelet transform is used as an example of a particular type of subband decomposition, the present invention can be adapted to other subband decomposition transforms or methods.

Method **900** starts in step **905** and proceeds to step **910**, where a rate distortion model is initialized. Specifically, method **900** employs a quadratic rate distortion model. The basic concept of the present invention is based on the strong correlation between the amplitudes of the hierarchical wavelet coefficients across several layers. In the Multiple-Quantization (MQ) mode of the current MPEG-4 Visual Texture Coding (VTC) tool, a different quantization scale can be applied to each individual wavelet spatial layer. The MQ provides more flexibility to improve the coding efficiency and perceptual image quality. The present technique can be extended to the other wavelet based still texture coding methods. Since the quantization scale is considered as a distortion measure, the quadratic rate distortion model for target bit-rate allocation can be described as

where R_{i}, is the total number of bits used for encoding the current wavelet spatial layer i, H_{i }is the bits used for header information, S_{i }is the encoding complexity in terms of quantization errors per wavelet coefficient in a wavelet spatial layer i, Q_{i }is the quantization scale used for the current wavelet spatial layer i, and A_{1 }and A_{2 }are the first order and the second order coefficients. In step **910**, the model of equation (1) is initialized, e.g., the desired target bits are distributed to each wavelet layer equally or proportionately to energy distribution, e.g., the square of the wavelet coefficients. Alternatively, an identical quantization parameter can be applied for all resolution layers. Namely, initial R_{i}, S_{i}, and Q_{i }are available via some preliminary computations and/or assumptions, e.g., use the same S_{i }from a previous frame, divide a target frame rate evenly for the number of spatial layers i to obtain R_{i }and the like.

In step **920**, the initial or initialized values are used to encode the various subband layers of the decomposed image signal. For example, Q_{i }is used to quantize a subband layer i, where a reasonable Q_{i }can be a value of 15 for some applications.

In step **930**, various model parameters of equation (1) can be updated based upon actual data resulting from encoding the image as discussed below. However, at the initial stage, there may not be sufficient data to begin updating the model parameters. Thus, step **930** can be skipped as shown by the dashed line.

In step **940**, method **900** queries whether the target rate is met. In other words, method **900** queries whether the initialized values were appropriately selected, e.g., whether “actual R_{i}” (actual_bitrate) is within a tolerance range to the initialized R_{i }(target_bitrate). Another way of viewing this query, is whether the sum of the R_{i }for all the layers is within a tolerance range to the target bitrate for the entire frame. If the query is affirmatively answered, then method **900** proceeds to step **990**. Namely, the initial values were adequate to meet the target bitrate calculated for the frame.

If the query is negatively answered, then method **900** proceeds to step **950**. Namely, the initial values were not adequate to meet the target bitrate calculated for the frame and refinement is necessary.

In step **950**, method **900** reallocate the target bitrate, i.e., a new R_{i }is computed for each subband layer. Specifically, the target bitrate is computed based on the available bits. Since each layer has an individual statistic characteristic with correlation to its neighboring layers, the bitrate for each spatial layer is estimated based on its encoding complexity and a weighting factor W. The encoding complexity is computed in terms of mean square quantization error (MSE). The weighting factor depends on the target bitrate and the weighted interaction factor dependent on the number of subband or spatial layers and forward predicted MSE. It should be understood that although the present invention discloses the use of MSE as a measure of quantization error, other measures, e.g., sum of absolute difference (SAD) and the like can be adapted into the present invention.

The basic principle of the bit allocation is to minimize the overall quantization distortion subject to a total bit budget constraint. The Lagragian multiplier is one approach to obtain such an optimal allocation for image and video coding. However, a potential drawback is its higher complexity with possibly negative bitrate allocation, where a convergence method is then necessary to remove such potential problem. Thus, the present invention employs a heuristic and efficient bit technique for each subband layer (wavelet or resolution layer) based on the mean square error (MSE) introduced by the quantization process with its associated weighting factor. Namely, the mean square error (MSE) serves as the basis for determining the complexity S_{i}.

However, since each wavelet decomposition step doubles its dynamic range, the MSE in resolution layer i will be normalized as MSE(i) defined in the following equation by the wavelet layer to avoid such a doubling factor:

In this equation, the symbol k is the number of wavelet decomposition layers, the symbol j denotes a spatial location in resolution layer i and the symbols >> and << are right and left bit shifting operations, respectively. It should be noted that NMSE(i) is equivalent to S_{i}.

Next, a weighting factor is computed. The weighting factor is used to prioritize the importance among resolution layers and to predict the next MSE for faster convergence. Although the normalized MSE is a good indicator for bit allocation, the lack of accuracy in predicting the RD relations for the next iteration needs to be considered. The weighting factor is determined a priori by the MSE of the next iteration in each individual resolution layer. Note that the MSE of the next iteration is an actual MSE since the quantization process is performed in the wavelet domain. With this accurate MSE, the weighting factor can be used to forward predict and heuristically reduce the overall MSE. It is important to note that this factor is generated automatically based on the characteristics of source images, the bit allocation among layers can be adaptively adjusted. The weighting factor calculation and bitrate allocation among wavelet layers are presented as follows:

Weighting Factor Calculation

if (actual_bitrate > target_bitrate) { |

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

layer_MSE[i] = compute_MSE(i, current_QP+1); |

total_MSE = total_MSE + layer_MSE[i];} |

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

weight[i] = layer_MSE[i]/total_MSE;} (3) |

else {/* actual_bitrate < target_bitrate */ |

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

layer_MSE[i] = compute_MSE(i, current_QP−1); |

total_MSE = total_MSE + layer_MSE[i];} |

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

weight[i] = layer_MSE[i]/total_MSE; |

} |

Target Bitrate Allocation

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

total_weightMSE = total_weightMSE + current_MSE[i] * |

weight[i]; |

} (4) |

for (i=0; i<wvt_decomp_layers; i++) { /* resolution layer i */ |

R[i] target_bitrate * current_MSE[i] * weight[i]/total_weightMSE; |

Using equation (4), a new target R_{i }is calculated for each layer. It should be noted that “target_bitrate” is the target bitrate for the entire frame and “current_MSE[i]” represents the actual MSE resulting from the encoding process and that “layer_MSE[i]” represents a MSE for the next iteration of encoding based upon a change in the quantization scale for that layer i.

In step **960**, method **900** computes a new quantization scale for each layer. Specifically, for each resolution layer i, a quantization step Q_{i }is derived based on the model parameters H_{i}, A_{1 }and A_{2 }and the recalculated S_{i }and R_{i }in accordance with equations (2) and (4).

In step **970**, method **900** employs the newly computed Q_{i }to encode the various subband layers of the decomposed image signal.

In step **980**, method **900** queries whether a threshold or a predefined number of refinement iterations has been achieved. Namely, the adjustment of the computed Q_{i }can be stopped after a number of iterations depending on a specific application, e.g., the available computational cycles or the time constraint for encoding the image sequence. If the query is affirmatively answered, then method **900** proceeds to step **990**. If the query is negatively answered, then method **900** proceeds to step **930**, where another iteration of adjustment is performed.

In step **990**, method **900** queries whether additional frames are to be encoded. If the query is affirmatively answered, then method **900** proceeds to step **910**, where the entire process is performed again for another frame of the image sequence. If the query is negatively answered, then method **900** ends in step **995**.

As discussed above, the parameters of the quadratic rate distortion model of equation (1) can be updated in step **930** of FIG. **9**. Specifically, the rate distortion model is updated based on the encoding results of the current iteration. The bits used for the header are deducted since it is not related to Q_{i}. The model parameters A_{1 }and A_{2 }are updated as follows. Given the number of texture bits, R_{i}, the quantization level, Q_{i}, and the encoding complexity, S_{i}, for the current and a specified number of past iterations, the model parameters A_{1 }and A_{2 }can be calculated using a linear regression technique. To obtain a more accurate R-D model for each resolution layer, additional statistical methods described below are performed:

1. Selection of Data Points:

The quality and quantity of the data set play very important roles in the accuracy of the model. As for the quantity of the data set, the more data points selected will make it more likely to result in an accurate model but with higher complexities. As for the quality of the data set, several objective indices such as MSE of the current data point can be used. By considering all these factors, a sliding window based data selection mechanism is developed.

2. Calculation of the Model Parameters A_{1 }and A_{2}:

Based on these two values, the theoretical target bitrate can be calculated for each data point within the sliding window obtained in step 1. For the selected data points, the quantization levels and the actual bitrates at previous time instants are collected by the encoder. Using the linear regression technique, the two model parameters A_{2 }and A_{1 }can be obtained as shown below.

where n is the number of selected past iterations, Q_{i }and R_{i }are the actual average quantization levels and actual bit counts in the past, respectively.

Removal of the Outliers from the Data Set:

After the new model parameters A_{1 }and A_{2 }are derived, the encoder optionally performs the refinement step to remove some statistical outlier data points. The removal of the outliers from the data set can diminish the impact of abnormal data points in the model update process. The outliers are defined in the statistical sense, as those data points whose difference between the actual bitrate and the target bitrate is larger than k standard deviation (e.g., k=1 as applied in experiments). The actual bitrate for each past frame is known and the target bitrate is recomputed based on the new model parameters obtained in the previous step, the actual bitrate R_{i }and the average quantization level Q_{i}. By applying the concept of this statistical technique, more representative data points are selected, and the final model parameters can be derived based on these new data points. Note that, to avoid the removal of all data points, the latest data point is always selected in the data set.

FIG. 10 illustrates an encoding system **1000** of the present invention. The encoding system comprises a general purpose computer **1010** and various input/output devices **1020**. The general purpose computer comprises a central processing unit (CPU) **1012**, a memory **1014** and an encoder **1016** for receiving and encoding a sequence of images.

In the preferred embodiment, the encoder **1016** is simply the encoder system **100** and/or encoder **106** as discussed above. The encoder **1016** can be a physical device which is coupled to the CPU **1012** through a communication channel. Alternatively, the encoder **1016** can be represented by a software application which is loaded from a storage device and resides in the memory **1012** of the computer. As such, the encoder **100** and **106** of the present invention can be stored on a computer readable medium.

The computer **1010** can be coupled to a plurality of input and output devices **1020**, such as a keyboard, a mouse, a camera, a camcorder, a video monitor, any number of imaging devices or storage devices, including but not limited to, a tape drive, a floppy drive, a hard disk drive or a compact disk drive. The input devices serve to provide inputs to the computer for producing the encoded video bitstreams or to receive the sequence of video images from a storage device or an imaging device.

Although various embodiments which incorporate the teachings of the present invention have been shown and described in detail herein, those skilled in the art can readily devise many other varied embodiments that still incorporate these teachings.

Patent Citations

Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US5333012 | Oct 14, 1992 | Jul 26, 1994 | Bell Communications Research, Inc. | Motion compensating coder employing an image coding control method |

US5528238 | Apr 28, 1994 | Jun 18, 1996 | Intel Corporation | Process, apparatus and system for decoding variable-length encoded signals |

US5606371 | Nov 29, 1994 | Feb 25, 1997 | U.S. Philips Corporation | Video signal coding with proportionally integrating quantization control |

US5764805 | Oct 23, 1996 | Jun 9, 1998 | David Sarnoff Research Center, Inc. | Low bit rate video encoder using overlapping block motion compensation and zerotree wavelet coding |

US5790196 | Feb 14, 1997 | Aug 4, 1998 | Mitsubishi Electric Information Technology Center America, Inc. | Adaptive video coding method |

US6023296 * | Jun 15, 1998 | Feb 8, 2000 | Sarnoff Corporation | Apparatus and method for object based rate control in a coding system |

US6049630 * | Oct 27, 1997 | Apr 11, 2000 | America Online, Inc. | Data compression using adaptive bit allocation and hybrid lossless entropy encoding |

US6055017 * | Mar 19, 1997 | Apr 25, 2000 | Matsushita Electric Industrial Co., Ltd. | Adaptive scanning technique for efficient wavelet video coding |

US6167087 * | Dec 2, 1997 | Dec 26, 2000 | Sony Corporation | Picture signal encoding method and apparatus and signal recording medium |

Non-Patent Citations

Reference | ||
---|---|---|

1 | Cheng P-Y. et al.: "Video coding using embedded wavelet packet transform and motion compensation", Visual Information Processing V, Orlando, FL, USA, Apr. 9, 1996; vol. 2753, pp. 64-75, XP000933680, Proceedings of the SPIE-The International Society for Optical Engineering, 1996, SPIE-Int. Soc. Opt. Eng. USA ISSN: 0277-786X, Sections 1,2.2, Figure 4. | |

2 | Cheng P-Y. et al.: "Video coding using embedded wavelet packet transform and motion compensation", Visual Information Processing V, Orlando, FL, USA, Apr. 9, 1996; vol. 2753, pp. 64-75, XP000933680, Proceedings of the SPIE—The International Society for Optical Engineering, 1996, SPIE-Int. Soc. Opt. Eng. USA ISSN: 0277-786X, Sections 1,2.2, Figure 4. | |

3 | Sodagar I. et al.: "Scalable Wavelet Coding for Synthetic/Natural Hybrid Images", IEEE Transactions on Circuits and Systems for Video Technology, US, IEEE Inc. New York, vol. 9, No. 2, Mar. 1999, pp. 244-254, XP000805529, ISSN: 1051-8215, Sections II.A-C, Figure 9. | |

4 | Xiong Z. et al: "Joint Optimization of Scalar and Tree-Structured Quantization of Wavelet Image Decompositions", Proceedings of the Asilomar Conference, US, NY, IEEE, Nov. 1, 1993, pp. 891-895, XP000438424, sections 2-3.1. | |

5 | Xiong Z. et al: "Joint Optimization of Scalar and Tree-Structured Quantization of Wavelet Image Decompositions", Proceedings of the Asilomar Conference, US, NY, IEEE, Nov. 1, 1993, pp. 891-895, XP0004348424, sections 2-3.1. |

Referenced by

Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US6532265 * | Mar 24, 2000 | Mar 11, 2003 | Imec Vzw | Method and system for video compression |

US6711300 * | Aug 15, 2002 | Mar 23, 2004 | Envivio, Inc. | Quantization and compression of information in a direct acyclic graph |

US6876770 * | Dec 7, 2001 | Apr 5, 2005 | Nikon Corporation | Image encoding device, electronic camera and recording medium for image encoding program |

US6891889 * | Sep 5, 2001 | May 10, 2005 | Intel Corporation | Signal to noise ratio optimization for video compression bit-rate control |

US6891975 * | Dec 21, 2001 | May 10, 2005 | Nikon Corporation | Image encoding device, image decoding device, electronic camera and recording medium |

US6940903 * | Mar 5, 2002 | Sep 6, 2005 | Intervideo, Inc. | Systems and methods for performing bit rate allocation for a video data stream |

US7020206 * | Nov 27, 2000 | Mar 28, 2006 | Intel Corporation | Wavelet coding of video |

US7039251 | Mar 22, 2004 | May 2, 2006 | Envivio, Inc. | Quantization and compression of information in a direct acyclic graph |

US7099513 * | Jun 18, 2002 | Aug 29, 2006 | Pts Corporation | Bit allocation process for multi-stage image compression |

US7209519 * | Apr 16, 2001 | Apr 24, 2007 | Mitsubishi Electric Research Laboratories, Inc. | Encoding a video with a variable frame-rate while minimizing total average distortion |

US7221706 * | Oct 4, 2004 | May 22, 2007 | Intervideo, Inc. | Systems and methods for performing bit rate allocation for a video data stream |

US7260826 * | May 25, 2001 | Aug 21, 2007 | Microsoft Corporation | Resource allocation in multi-stream IP network for optimized quality of service |

US7277588 | Apr 18, 2006 | Oct 2, 2007 | Envivio, Inc. | Quantization and compression of information in a direct acyclic graph |

US7313814 * | Apr 1, 2003 | Dec 25, 2007 | Microsoft Corporation | Scalable, error resilient DRM for scalable media |

US7406176 | Apr 1, 2003 | Jul 29, 2008 | Microsoft Corporation | Fully scalable encryption for scalable multimedia |

US7450571 | Jan 27, 2005 | Nov 11, 2008 | Microsoft Corporation | Resource allocation in multi-stream IP network for optimized quality of service |

US7474796 | Oct 2, 2007 | Jan 6, 2009 | Envivio, Inc | Quantization and compression of information in a direct acyclic graph |

US7522724 * | Jan 7, 2005 | Apr 21, 2009 | Hewlett-Packard Development Company, L.P. | System and method of transmission of generalized scalable bit-streams |

US7559078 * | Jul 7, 2009 | Microsoft Corporation | Resource allocation in multi-stream IP network for optimized quality of service | |

US7574061 * | Sep 23, 2002 | Aug 11, 2009 | Ricoh Company, Ltd. | Encoding device and method utilizing table data for selection of bit planes |

US7574726 | Mar 23, 2006 | Aug 11, 2009 | Microsoft Corporation | Resource allocation in multi-stream IP network for optimized quality of service |

US7639886 * | Dec 29, 2009 | Adobe Systems Incorporated | Determining scalar quantizers for a signal based on a target distortion | |

US7653255 | Jan 26, 2010 | Adobe Systems Incorporated | Image region of interest encoding | |

US7697608 | Nov 19, 2004 | Apr 13, 2010 | Sony Corporation | Scalable MPEG video/macro block rate control |

US7756271 * | Jun 15, 2004 | Jul 13, 2010 | Microsoft Corporation | Scalable layered access control for multimedia |

US7773672 * | May 30, 2006 | Aug 10, 2010 | Freescale Semiconductor, Inc. | Scalable rate control system for a video encoder |

US7822282 * | Mar 11, 2005 | Oct 26, 2010 | Canon Kabushiki Kaisha | Moving image coding apparatus, moving image decoding apparatus, control method therefor, computer program, and computer-readable storage medium |

US7835437 * | Aug 1, 2003 | Nov 16, 2010 | Ji Zhang | Statistical remultiplexing of compressed video segments |

US7916952 * | Sep 12, 2005 | Mar 29, 2011 | Gary Demos | High quality wide-range multi-layer image compression coding system |

US7957604 | Mar 11, 2005 | Jun 7, 2011 | Canon Kabushiki Kaisha | Moving image coding apparatus, moving image decoding apparatus, control method therefor, and computer-readable storage medium |

US8027384 | Sep 27, 2011 | Sony Corporation | Scalable MPEG video/macro block rate control | |

US8077775 | Dec 13, 2011 | Freescale Semiconductor, Inc. | System and method of adaptive rate control for a video encoder | |

US8090730 * | Dec 4, 2002 | Jan 3, 2012 | University Of Southern California | Methods for fast progressive evaluation of polynomial range-sum queries on real-time datacubes |

US8218627 | Feb 25, 2010 | Jul 10, 2012 | Sony Corporation | Scalable MPEG video/macro block rate control |

US8270474 | Jun 1, 2011 | Sep 18, 2012 | Samsung Electronics Co., Ltd. | Image encoding and decoding apparatus and method |

US8345748 | Jun 1, 2011 | Jan 1, 2013 | Samsung Electronics Co., Ltd. | Image encoding and decoding apparatus and method |

US8401315 | Mar 19, 2013 | Gary Demos | File and folder structure for layered compressed image sequences | |

US8428364 | Jan 7, 2011 | Apr 23, 2013 | Dolby Laboratories Licensing Corporation | Edge enhancement for temporal scaling with metadata |

US8437389 | Oct 12, 2010 | May 7, 2013 | Software Site Applications, Limited Liability Company | Statistical remultiplexing of compressed video segments |

US8666184 | Feb 20, 2013 | Mar 4, 2014 | Gary Demos | Variable length coding system |

US8897370 * | Nov 30, 2009 | Nov 25, 2014 | Google Inc. | Bitrate video transcoding based on video coding complexity estimation |

US9042442 * | Jul 19, 2004 | May 26, 2015 | Samsung Electronics Co., Ltd. | Image encoding and decoding apparatus and method |

US9042443 | Nov 10, 2014 | May 26, 2015 | Samsung Electronics Co., Ltd. | Image encoding and decoding apparatus and method |

US9185412 | Feb 25, 2014 | Nov 10, 2015 | Gary Demos | Signal to noise improvement |

US9371099 | Jul 19, 2011 | Jun 21, 2016 | The Wilfred J. and Louisette G. Lagassey Irrevocable Trust | Modular intelligent transportation system |

US20020010938 * | May 25, 2001 | Jan 24, 2002 | Qian Zhang | Resource allocation in multi-stream IP network for optimized quality of service |

US20020081036 * | Dec 21, 2001 | Jun 27, 2002 | Nikon Corporation | Image encoding device, image decoding device, electronic camera and recording medium |

US20020122489 * | Dec 7, 2001 | Sep 5, 2002 | Nikon Corporation | Image encoding device, electronic camera and recording medium for image encoding program |

US20030007558 * | Apr 16, 2001 | Jan 9, 2003 | Mitsubishi Electric Research Laboratories, Inc. | Encoding a video with a variable frame-rate while minimizing total average distortion |

US20030067981 * | Mar 5, 2002 | Apr 10, 2003 | Lifeng Zhao | Systems and methods for performing bit rate allocation for a video data stream |

US20030072364 * | Sep 5, 2001 | Apr 17, 2003 | Intel Corporation | Signal to noise ratio optimization for video compression bit-rate control |

US20030081847 * | Sep 23, 2002 | May 1, 2003 | Yutaka Sato | Encoding device and method |

US20030123539 * | Dec 28, 2001 | Jul 3, 2003 | Hyung-Suk Kim | Method and apparatus for video bit-rate control |

US20030204499 * | Dec 4, 2002 | Oct 30, 2003 | Cyrus Shahabi | Methods for fast progressive evaluation of polynomial range-sum queries on real-time datacubes |

US20030231797 * | Jun 18, 2002 | Dec 18, 2003 | Pulsent Corporation, A Corporation Of The State Of California | Bit allocation process for multi-stage image compression |

US20040032422 * | Aug 15, 2002 | Feb 19, 2004 | Envivio.Com. Inc | Quantization and compression of information in a direct acyclic graph |

US20040196972 * | Apr 1, 2003 | Oct 7, 2004 | Bin Zhu | Scalable, error resilient DRM for scalable media |

US20040196975 * | Apr 1, 2003 | Oct 7, 2004 | Microsoft Corporation | Fully scalable encryption for scalable multimedia |

US20040252904 * | Mar 22, 2004 | Dec 16, 2004 | Envivio, Inc. | Quantization and compression of information in a direct acyclic graph |

US20050013360 * | Jul 19, 2004 | Jan 20, 2005 | Samsung Electronics Co., Ltd. | Image encoding and decoding apparatus and method |

US20050047509 * | Aug 25, 2004 | Mar 3, 2005 | Samsung Electronics Co., Ltd. | Scalable video coding and decoding methods, and scalable video encoder and decoder |

US20050058199 * | Oct 4, 2004 | Mar 17, 2005 | Lifeng Zhao | Systems and methods for performing bit rate allocation for a video data stream |

US20050141608 * | Sep 8, 2004 | Jun 30, 2005 | Samsung Electronics Co., Ltd. | Pipeline-type operation method for a video processing apparatus and bit rate control method using the same |

US20050155080 * | Jan 27, 2005 | Jul 14, 2005 | Microsoft Corporation | Resource allocation in multi-stream IP network for optimized quality of service |

US20050169369 * | Sep 17, 2004 | Aug 4, 2005 | Sony Corporation | Scalable MPEG video/macro block rate control |

US20050169370 * | Nov 19, 2004 | Aug 4, 2005 | Sony Electronics Inc. | Scalable MPEG video/macro block rate control |

US20050214428 * | Mar 23, 2004 | Sep 29, 2005 | Yonas Gizaw | Non-steeped corn blend and article of commerce |

US20050276416 * | Jun 15, 2004 | Dec 15, 2005 | Microsoft Corporation | Scalable layered access control for multimedia |

US20060071825 * | Sep 12, 2005 | Apr 6, 2006 | Gary Demos | High quality wide-range multi-layer image compression coding system |

US20060153373 * | Jan 7, 2005 | Jul 13, 2006 | Debargha Mukherjee | System and method of transmission of generalized scalable bit-streams |

US20060156200 * | Mar 23, 2006 | Jul 13, 2006 | Microsoft Corporation | Resource Allocation in Multi-Stream IP Network for Optimized Quality of Service |

US20060156201 * | Mar 23, 2006 | Jul 13, 2006 | Microsoft Corporation | Resource Allocation in Multi-Stream IP Network for Optimized Quality of Service |

US20060245659 * | Apr 18, 2006 | Nov 2, 2006 | Envivio, Inc. | Quantization and compression of information in a direct acyclic graph |

US20070160299 * | Mar 11, 2005 | Jul 12, 2007 | Canon Kabushiki Kaisha | Moving image coding apparatus, moving image decoding apparatus, control method therefor, computer program, and computer-readable storage medium |

US20070263720 * | May 12, 2006 | Nov 15, 2007 | Freescale Semiconductor Inc. | System and method of adaptive rate control for a video encoder |

US20070280349 * | May 30, 2006 | Dec 6, 2007 | Freescale Semiconductor Inc. | Scalable rate control system for a video encoder |

US20080056354 * | Aug 29, 2006 | Mar 6, 2008 | Microsoft Corporation | Transcoding Hierarchical B-Frames with Rate-Distortion Optimization in the DCT Domain |

US20080219571 * | Mar 11, 2005 | Sep 11, 2008 | Masaki Suzuki | Moving Image Coding Apparatus, Moving Image Decoding Apparatus, Control Method Therefor, Computer Program, and Computer-Readable Storage Medium |

US20100150227 * | Feb 25, 2010 | Jun 17, 2010 | Sony Corporation | Scalable mpeg video/macro block rate control |

US20100150228 * | Feb 25, 2010 | Jun 17, 2010 | Sony Corporation | Scalable mpeg video/macro block rate control |

US20110026584 * | Feb 3, 2011 | Ji Zhang | Statistical remultiplexing of compressed video segments | |

US20110032994 * | Jan 3, 2008 | Feb 10, 2011 | Numeri Ltd. | method and a system for wavelet based processing |

US20110194600 * | Aug 11, 2011 | Gary Demos | File and folder structure for layered compressed image sequences | |

US20110235710 * | Sep 29, 2011 | Samsung Electronics Ltd., Co. | Image encoding and decoding apparatus and method | |

CN100559878C * | Jul 7, 2004 | Nov 11, 2009 | 阿尔卡特公司 | Wavelet transformation method and apparatus for compressing digital image data |

CN100591121C | Apr 1, 2004 | Feb 17, 2010 | 微软公司 | Upgradable error elastic DRM used for upgradable medium |

EP1465426A1 * | Mar 18, 2004 | Oct 6, 2004 | Microsoft Corporation | Scalable and error resilient digital rights management (DRM) for scalable media |

Classifications

U.S. Classification | 375/240.03, 382/251, 375/E07.051, 375/E07.072, 382/240, 375/E07.036, 382/252, 375/E07.153, 382/248, 382/239, 375/E07.05 |

International Classification | H04N7/26 |

Cooperative Classification | H04N19/647, H04N19/115, H04N19/146, H04N19/147, H04N19/10, H04N19/63, H04N19/61 |

European Classification | H04N7/26H50E5A, H04N7/26H30E5A, H04N7/26A6D, H04N7/26H30H6, H04N7/26H30C |

Legal Events

Date | Code | Event | Description |
---|---|---|---|

Nov 8, 2000 | AS | Assignment | Owner name: SARNOFF CORPORATION, NEW JERSEY Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:LEE, HUNG-JU;SODAGAR, IRAJ;ZHANG, YA-QIN;AND OTHERS;REEL/FRAME:011262/0858;SIGNING DATES FROM 20001005 TO 20001101 |

Aug 26, 2005 | FPAY | Fee payment | Year of fee payment: 4 |

Aug 26, 2009 | FPAY | Fee payment | Year of fee payment: 8 |

Oct 4, 2013 | REMI | Maintenance fee reminder mailed | |

Feb 26, 2014 | LAPS | Lapse for failure to pay maintenance fees | |

Apr 15, 2014 | FP | Expired due to failure to pay maintenance fee | Effective date: 20140226 |

Rotate